The Cost of Care: Is There an Optimal Level of Expenditure?
McGuire, Alistair, Serra, Victoria, Harvard International Review
Many countries' concern over their level of health care expenditure raises the question of what the optimal level of national expenditure devoted to health care ought to be. It is, unfortunately, extremely difficult to define such a universal level of health care expenditure. A basic problem is that the objectives of the health care sector are not normally defined in an explicit manner. Efficiency is often stated as an objective, but is seldom quantified. Common wisdom holds equity as a major consideration in the resource allocation process, but equity, too, is hardly ever explicitly defined. Yet until the objectives are clearly outlined, it is impossible to state whether health care is under- or over-funded. Many arguments on both sides are misleading, generally supporting long-held biases or confusing the relevant issues. An economic perspective can, however, illuminate some points of this discussion.
The past 30 years have experienced a slight but steady increase in the share of national resources devoted to health care in a number of developed countries. This trend indicates that the income elasticity of health care expenditures, defined as the proportionate change in expenditure given a proportionate change in national income, is greater than one--a one percent increase in gross domestic product (GDP) is associated with a greater than one percent increase in health care spending.
Real Health Care Gains
That a number of countries are indeed devoting more national income to health care each year is shown in Figure 1 of "Health Care Headaches," which reports data on a few major Organization for Economic Cooperation and Development (OECD) countries. Even during periods in which gross domestic product (GDP) growth rates are relatively low, growth in health care expenditure remains high. There are, of course, some periods of exception to this rule, but the general trend holds. Indeed, since 2000, growth rates in health care expenditure have far outstripped growth in GDP for most of these countries. Although the period is smaller and the choice of a base year therefore probably matters more, these relatively high growths in health expenditure occurred when GDP growth rates were relatively modest.
It could be suggested that these recent growth rates are restoring the relatively high rates of expenditure of the 1960s and early 1970s. But given that health care expenditure is at a much higher level now than it was in those earlier periods, these higher growths in expenditure do in fact appear exceptional. Reliance on the restoration of historical rates of growth as the sole justification for increased expenditure has little real defense as an argument for retaining high health care expenditure.
Another equally simple, supposed justification for the increased levels of health care expenditure compares relative expenditure across different countries by regressing health care expenditure per capita on GDP per capita. Such a regression is typically used to show that GDP per capita "explains" a remarkable 90 percent of the variation in health care expenditure per head across the majority of developed countries, as shown in Figure 2 of "Health Care Headaches." This explanation of the pattern of health care expenditure fails, however, when the United States and Luxemburg, two outliers, are taken into consideration. Therefore, such regression does not necessarily reflect any causal relationship between health and wealth at an aggregate level; high levels of GDP do not necessarily lead to high levels of health care expenditure. Nor can such relationships be used to determine whether some countries are under- or over-spending on health care.
Regression operates on the basis of random variation. In other words, some observations will lie below and some above the regression line. The interpretation that because an observation lies below or above the regression line, it is evidence of under-spending or over-spending, involves a value judgment that observations that lie on the regression line are somehow "appropriate. …